WIP

src/common/loadable.rs

@@ -3,7 +3,7 @@ use std::iter::once;
 
 use crate::error::Error;
 use crate::{
-    common::{Relocatable, Section, SectionInfo, SymbolIndex},
+    common::{Relocatable, Section, SectionInfo, SymbolIndex, Symbols},
     error::LinkError,
 };
 
@@ -42,7 +42,7 @@ pub struct Loadable {
     rodata: SegmentSections,
     data: SegmentSections,
     bss: SegmentSections,
-    symbol_map: HashMap<String, Vec<SymbolIndex>>,
+    symbol_map: HashMap<String, Symbols>,
 }
 
 impl Loadable {
@@ -57,30 +57,36 @@ impl Loadable {
         Ok(())
     }
 
-    pub fn process_symbol<R>(&mut self, index: usize, object: &R) -> Result<(), Error>
+    pub fn process_symbol<R>(&mut self, si: SymbolIndex, objects: &[R]) -> Result<(), Error>
     where
         R: Relocatable,
     {
-        if !object.symbol_needs_resolving(index) {
+        let object = objects
+            .get(si.object_index)
+            .ok_or(Error::InvalidSectionIndex)?;
+
+        if !object.symbol_needs_resolving(si.symbol_index) {
             return Ok(()); // local symbols are not stored
         }
 
-        let name = object.symbol_name(index)?;
+        let name = object.symbol_name(si.symbol_index)?;
-        eprintln!("NAME: {}", name);
+        eprintln!("NAME: {}/{:?}", name, object.symbol_value(si.symbol_index)?);
 
         let si = SymbolIndex {
             object_index: object.object_index(),
-            symbol_index: index,
+            symbol_index: si.symbol_index,
         };
 
         if let Some(existing) = self.symbol_map.get_mut(name) {
-            existing.push(si);
+            existing.push(si, objects)
         } else {
-            self.symbol_map.insert(name.to_owned(), vec![si]);
+            let mut symbols = Symbols::default();
-        }
+            symbols.push(si, objects)?;
+            self.symbol_map.insert(name.to_owned(), symbols);
 
             Ok(())
         }
+    }
 
     pub fn segment_sections(&self) -> impl Iterator<Item = &SegmentSections> {
         let text = once(&self.text);

src/common/relocatable.rs

@@ -23,6 +23,10 @@ pub trait Relocatable: Display + Sized {
 
     fn symbol_name(&self, index: usize) -> Result<&str, Error>;
 
+    fn symbol_value(&self, index: usize) -> Result<&[u8], Error>;
+
+    fn symbol_file_offset(&self, index: usize) -> Result<usize, Error>;
+
     fn symbol_needs_resolving(&self, index: usize) -> bool;
 
     fn bytes(&self, offset: u64, size: u64) -> Result<&[u8], Error>;

src/common/symbol.rs

@@ -1,7 +1,92 @@
 use std::fmt::Debug;
 
-#[derive(Default, Debug)]
+use crate::error::{Error, LinkError, Trace, LE_GLOBAL_SYMBOL_DUPLICATE};
+
+use super::Relocatable;
+
+#[derive(Default, Debug, Clone, Copy, PartialEq, Eq)]
 pub struct SymbolIndex {
     pub object_index: usize,
     pub symbol_index: usize,
 }
+
+#[derive(Default)]
+pub struct Symbols {
+    indexes: Vec<SymbolIndex>,
+    value_index: Option<usize>,
+}
+
+impl Symbols {
+    pub fn push<R>(&mut self, si: SymbolIndex, objects: &[R]) -> Result<(), Error>
+    where
+        R: Relocatable,
+    {
+        let object = objects
+            .get(si.object_index)
+            .ok_or(Error::InvalidSymbolIndex)?;
+        let value = object.symbol_value(si.symbol_index)?;
+
+        self.indexes.push(si);
+
+        if value.len() > 0 {
+            if let Some(existing) = self.value_index {
+                let osi = self
+                    .indexes
+                    .get(existing)
+                    .ok_or(Error::InvalidSymbolIndex)?;
+                Self::duplicate_symbol_error(si, *osi, objects)
+            } else {
+                self.value_index = Some(self.indexes.len() - 1);
+
+                Ok(())
+            }
+        } else {
+            Ok(())
+        }
+    }
+
+    pub fn value<'o, R>(&self, object: &'o R) -> Result<&'o [u8], Error>
+    where
+        R: Relocatable,
+    {
+        self.value_index.map_or(Ok(&[]), |v| {
+            object.symbol_value(self.indexes[v].symbol_index)
+        })
+    }
+
+    fn duplicate_symbol_error<R>(
+        si: SymbolIndex,
+        osi: SymbolIndex,
+        objects: &[R],
+    ) -> Result<(), Error>
+    where
+        R: Relocatable,
+    {
+        let object = objects
+            .get(si.object_index)
+            .ok_or(Error::InvalidSymbolIndex)?;
+        let other = objects
+            .get(osi.object_index)
+            .ok_or(Error::InvalidObjectIndex)?;
+        let mut traces = Vec::new();
+
+        traces.push(Trace {
+            origin: object.origin().into(),
+            offset: object.symbol_file_offset(si.symbol_index)?,
+            source_info: None,
+        });
+        traces.push(Trace {
+            origin: other.origin().into(),
+            offset: other.symbol_file_offset(osi.symbol_index)?,
+            source_info: None,
+        });
+
+        let link_error = LinkError {
+            code: LE_GLOBAL_SYMBOL_DUPLICATE,
+            message: "duplicate global symbol found".into(),
+            traces,
+        };
+
+        Err(Error::LinkingError(link_error))
+    }
+}

src/error.rs

@@ -9,7 +9,9 @@ pub const LE_GLOBAL_SYMBOL_DUPLICATE: u32 = 10001;
 pub enum Error {
     IOError(Box<dyn std::error::Error>),
     DataError(PathBuf), // generic "something wrong with data" error, not specific to parsing
+    InvalidObjectIndex,
     InvalidObjectType(u32),
+    InvalidSectionIndex,
     InvalidSectionName,
     InvalidSymbolIndex,
     InvalidSymbolName,
@@ -23,7 +25,7 @@ pub enum Error {
 #[derive(Debug, Default)]
 pub struct Trace {
     pub origin: PathBuf,
-    pub offset: u64, // 0 indicates unknown/invalid
+    pub offset: usize, // 0 indicates unknown/invalid
     pub source_info: Option<Box<dyn SourceInfo>>,
 }
 
@@ -41,11 +43,13 @@ impl Error {
         match self {
             Error::IOError(_) => 100,
             Error::DataError(_) => 800,
-            Error::InvalidObjectType(_) => 200,
+            Error::InvalidObjectIndex => 200,
-            Error::InvalidSectionName => 300,
+            Error::InvalidObjectType(_) => 201,
-            Error::InvalidSymbolIndex => 325,
+            Error::InvalidSectionIndex => 300,
-            Error::InvalidSymbolName => 350,
+            Error::InvalidSectionName => 301,
-            Error::InvalidSectionData => 400,
+            Error::InvalidSectionData => 302,
+            Error::InvalidSymbolIndex => 400,
+            Error::InvalidSymbolName => 401,
             Error::ParseError(_) => 500,
             Error::TryFromIntError => 600,
             Error::InvalidInput => 700,
@@ -79,11 +83,13 @@ impl Display for Error {
         match self {
             Error::IOError(err) => write!(f, "IO error: {}", err),
             Error::DataError(origin) => write!(f, "Data error in {:?}", origin),
+            Error::InvalidObjectIndex => write!(f, "Invalid section index"),
             Error::InvalidObjectType(ot) => write!(f, "Invalid object type: {}", ot),
+            Error::InvalidSectionIndex => write!(f, "Invalid section index"),
             Error::InvalidSectionName => write!(f, "Invalid section name"),
+            Error::InvalidSectionData => write!(f, "Invalid section data"),
             Error::InvalidSymbolIndex => write!(f, "Invalid symbol index"),
             Error::InvalidSymbolName => write!(f, "Invalid symbol name"),
-            Error::InvalidSectionData => write!(f, "Invalid section data"),
             Error::ParseError(err) => write!(f, "Parse error: {}", err),
             Error::TryFromIntError => write!(f, "Integer conversion error"),
             Error::InvalidInput => write!(f, "Input object invalid"),

src/formats/elf/object.rs

@@ -48,6 +48,23 @@ impl ElfObject {
         Ok(result)
     }
 
+    fn resolve_symbol_value(&self, symbol_index: usize) -> Result<&[u8], Error> {
+        if let Some((_, s64)) = self.symbols.get(symbol_index) {
+            if s64.st_size == 0 {
+                Ok(&[])
+            } else if let Some(shdr) = self.shdrs.get(s64.st_shndx as usize) {
+                let start: usize = (shdr.sh_offset + s64.st_value) as usize;
+                let end: usize = start + s64.st_size as usize;
+
+                Ok(&self.data[start..end])
+            } else {
+                Err(Error::InvalidSectionIndex)
+            }
+        } else {
+            Err(Error::InvalidSymbolIndex)
+        }
+    }
+
     fn resolve_name(&self, name_idx: usize, sh_index: usize) -> Result<&str, Error> {
         if name_idx == 0 {
             let error = if sh_index == self.ehdr.e_shstrndx.into() {
@@ -194,6 +211,22 @@ impl Relocatable for ElfObject {
         }
     }
 
+    fn symbol_value(&self, index: usize) -> Result<&[u8], Error> {
+        self.resolve_symbol_value(index)
+    }
+
+    fn symbol_file_offset(&self, index: usize) -> Result<usize, Error> {
+        let (sh_index, s64) = self.symbols.get(index).ok_or(Error::InvalidSymbolIndex)?;
+
+        let shdr = self
+            .shdrs
+            .get(*sh_index)
+            .ok_or(Error::InvalidSectionIndex)?;
+        let sh_offset: usize = shdr.sh_offset as usize;
+
+        Ok(sh_offset)
+    }
+
     fn symbol_needs_resolving(&self, index: usize) -> bool {
         match self.symbols[index].1.get_bind() {
             Bind::Global | Bind::Weak => true,

src/linker.rs

@@ -10,7 +10,7 @@ where
     R: Relocatable,
     O: Output<R>,
 {
-    relocatables: Vec<R>,
+    objects: Vec<R>,
     loadable: Loadable,
     output: O,
 }
@@ -22,21 +22,21 @@ where
 {
     pub fn new(output: O) -> Self {
         Self {
-            relocatables: Vec::new(),
+            objects: Vec::new(),
             loadable: Loadable::default(),
             output,
         }
     }
 
     pub fn add_relocatable(&mut self, origin: PathBuf) -> Result<(), Error> {
-        let r = R::new(origin, self.relocatables.len())?;
+        let r = R::new(origin, self.objects.len())?;
 
         // TODO: parallelize?
         for section in r.sections() {
             self.loadable.process_section(section?)?;
         }
 
-        self.relocatables.push(r);
+        self.objects.push(r);
 
         Ok(())
     }
@@ -45,15 +45,19 @@ where
         self.process_symbols()?;
 
         self.loadable.set_start_offset(4096); // TODO: get from .start symbol location
-        self.output.finalize(&self.relocatables, &self.loadable)
+        self.output.finalize(&self.objects, &self.loadable)
     }
 
     fn process_symbols(&mut self) -> Result<(), Error> {
-        for r in &self.relocatables {
+        for r in &self.objects {
             let symbol_count = r.symbol_count();
 
             for i in 0..symbol_count {
-                self.loadable.process_symbol(i, r)?;
+                let si = SymbolIndex {
+                    object_index: r.object_index(),
+                    symbol_index: i,
+                };
+                self.loadable.process_symbol(si, &self.objects)?;
             }
         }
 
@@ -69,7 +73,7 @@ where
     fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
         writeln!(f, "===Relocatables===")?;
 
-        for r in self.relocatables.iter() {
+        for r in self.objects.iter() {
             write!(f, "{}", r)?;
         }